Skip to content
LyraShield AIOpen beta

For coding agents

Release assurance your coding agent can act on.

Give your coding agent evidence-backed checks, reviewable fix proposals, and a fresh retest before you ship.

Set up LyraShield for your agent

Six-stage release loop

  1. 01

    Target

  2. 02

    Review

  3. 03

    Evidence

  4. 04

    Fix proposal

  5. 05

    Retest

  6. 06

    Report

Two-command setup

npx lyrashield login --oauth
npx lyrashield init

Log in with OAuth first, then run the init command to detect and configure your agent.

  1. 1Sign in with OAuth — run npx lyrashield login --oauth to select a workspace and save the local credential.
  2. 2Install the Agent Plugin — run npx lyrashield init to detect and configure your coding agent.
  3. 3Choose your supported coding agent — use the matching integration guide for your client.
  4. 4Use review tools with approval — read-only tools are available after authentication; mutations require write scope and explicit human approval.

Supported coding agents (26)

npx lyrashield init detects which of these you have installed and configures each one. The badge says what setup looks like for that client; every name links to its own guide.

Agent Plugin · 5

Writes a config file · 13

Uses the agent's own CLI · 1

Shows values to paste · 7

What an agent can and cannot do

  • Read-only tools are available after workspace authentication.
  • Fixes are proposals for review, not automatic code changes or merges.
  • Mutating tools require write scope and explicit human approval outside the agent.

Read the Agent Plugin guide or the Remote MCP docs for the full setup and safety model.